Units with Hardware Issues

Initial Triage

Every unit that was dim after unplugging from the charger was recovered and analyzed.

• Unit 29 - passes all tests, no detectable errors. However, nofun mode was set.
• Unit 34 - passes all tests, no detectable errors.
• Unit 19 - passes all tests, no detectable errors.
• Unit 32 - refurb unit. 2x bad LEDs, right of LC2 and LC28. No obvious mechanical damage to LEDs.
• Unit 25 - LEDs go funky in dB color mode. LEDs OK with solid colors. Definitely needs more analysis.
• Unit 13 - passes all tests, no detectable errors. Definitely had trouble on-playa tho.
• Unit 24 - passes all tests, no detectable errors.
• Unit 36 - passes all tests, no detectable errors. Note on label says "dim"
• Unit 12 - passes all tests, no detectable errors.
• Unit 50 - passes all tests, no detectable errors.
• Unit 46 - passes all tests, no detectable errors.
• Unit 51 - passes all tests, no detectable errors.

Initial conclusions

At least 3-4 units that were recovered were definitely in a "reboot to red light" mode -- hitting the reset button would bring the board back, but it would display just a red light. Because none of the boards exhibited this issue upon examination, here are the top theories as to what may have happened:

• Some modules had a loose screw rattling around on the inside. The loose screw got the board into a bad state, which was unrecoverable somehow with a hard reset.
• Perhaps boards would only fail with the pressure of the battery pressing against the PCB, causing flexure and failure of solder joints
• We can rule out screws biting into the traces and causing failures, the boards were inspected for this, and while in a couple spots screws were seen biting into a ground plane none were seen cutting into traces
• The battery module itself being bad (so the battery is no longer taking charge -- for example, the battery consists of two cells joined together, and possibly the connection between them was weakened by vibration).
• Cabling going flakey -- in one case, this was definitely the situation, the charging cable had gotten caught in the lens-module screw threads and the negative wire got chewed up. However, the cable endings provided by the lens module maker are of low quality and could have high tolerances which would cause them to work their way out with vibration.
• Dusty charge port.

Suggestions to improve include:

• Properly sized battery pack that fits between the standoffs
• Metal bracket to hold battery in place underneath the PCB. This would avoid putting pressure from the battery directly against the PCB
• Blow duster gas into charge port before plugging in.
• Better charging-enabled UI so we know for sure if a cube is plugged in, even in bright sunlight
• No soldermask where the screws engage the PCB
• Improved cabling -- move connectors to a better location, and a cable format that is more reliable against high vibration situations (will be much more expensive, but look into automotive-rated solutions)